Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    RSI-BOOT

    1

    Stop your AI coding agent from repeating the same mistakes.

    Free

    Security scanned
    RSI-BOOT

    RSI-BOOT

    Example session with this skill installed

    I just fixed the recurring timeout issue in the auth middleware. Recall our conventions for this project and then help me distill this lesson so we don't break it again.

    • Read your context and instructions
    • Compiled the rsi-boot

    I have recalled the project memory and initialized the distillation process for the auth middleware fix.

    Project Memory & Distillation Report

    ActionTool UsedResult
    Recalled Constraintsrsi_recallIdentified 2 prohibitions: No external sync calls in middleware; No hardcoded timeout values.
    Knowledge Searchrsi_knowledge_searchFound related lesson from 2023-11-12 regarding JWT verification latency.
    Lesson Distilledrsi_learnCaptured wrong_action: Using default setTimeout without clearing; correct_fix: Implemented AbortController signal.
    Statusrsi_memoryNew lesson added to .rsi/ as Pending.
    Impactrsi_statsDistillation successful. Estimated 450 tokens saved for future related tasks.

    Next steps

    1. Run rsi_review to approve and inject this new lesson into the active project constraints.
    2. Use rsi_impact to view the updated health and efficiency metrics for this repository's memory.
    3. Call rsi_explore if you want to see how this fix relates to other middleware components in the graph.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Stop your AI coding agent from repeating the same mistakes. RSI Boot provides a local, zero-API-key memory layer that lives directly in your repository. It enables agents to recall project-specific conventions and prohibitions before starting a task and distill new lessons after completing one. By storing all data in a local .rsi/ folder, it ensures project intelligence stays on your machine without external network calls.

    What it does

    • Persistent memory stores project-specific rules, prohibitions, and past repair lessons locally.
    • Contextual recall surfaces relevant conventions and constraints at the start of every task.
    • Lesson distillation captures verified learnings and bug fixes to prevent future regressions.
    • Rule injection gating requires a formal review process before new lessons become active constraints.
    • Conflict detection identifies contradictory memories to maintain a clean and accurate knowledge base.

    How it works

    1. Recall: Run rsi_recall and rsi_explore at the start of a task to pull hard constraints and prohibitions.
    2. Execute: Perform the coding task following the surfaced project conventions.
    3. Distill: Use rsi_learn after fixing bugs or receiving corrections to document the error and the fix.
    4. Review: Execute rsi_review to approve pending lessons, promoting them into the active rule set.

    Frameworks & tools

    This skill requires the rsi-boot MCP server, Python 3.10+, and the rsi package installed via pip. It manages data within a local .rsi/ directory in your project root.

    Why this beats prompting it yourself

    Standard system prompts are static and consume significant token overhead. This skill dynamically injects only the relevant project history and rules, ensuring the agent learns from its specific mistakes in your codebase without manual prompt updates.

    Use cases

    • Preventing an agent from using deprecated libraries or banned design patterns.
    • Remembering specific architectural quirks that caused previous production bugs.
    • Ensuring consistent naming conventions across a large, multi-developer repository.
    • Measuring agent efficiency and token savings through local memory impact stats.

    Known limitations

    Requires the rsi MCP server to be configured and running. It does not perform business code edits during the memory audit phase.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.

    Requires Python 3.10+ and the rsi-boot MCP server. Memory is isolated per project in the .rsi/ folder. Business code edits are gated until memory audits finish.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 22 days ago

    • Free with an Agensi account

    Software
    PythonPublisher MCPuv (Python tool runner)Install Dark MCPInstall MCPGitjestNode.js

    Permissions

    Read Files
    Write Files
    Environment Variables
    Listed28 days ago
    Updated22 days ago

    What's inside

    Frequently Asked Questions

    RSI Boot is a local-first memory layer that prevents AI agents from repeating the same coding mistakes by storing project conventions and past bug fixes directly in your repository's `.rsi/` folder.

    The skill is compatible with any AI agent or IDE that supports the Model Context Protocol (MCP), provided you have Python 3.10+ and the `rsi` package installed on your local machine.

    Unlike static prompts, RSI Boot dynamically injects only the relevant project history for the current task, saving token costs and ensuring your agent follows specific, verified lessons rather than generic coding patterns.

    Your data remains entirely local; all memory, lessons, and conventions are stored in a `.rsi/` directory within your project root, meaning no external network calls or API keys are required.

    New learnings must pass through the `rsi_review` command, allowing you to approve or reject lessons before they are promoted to active constraints, ensuring your agent's knowledge base stays clean and accurate.

    New in LLM Ops

    Free